Details
-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
None
-
None
-
Undefined
-
---
-
---
Description
Two MP Config related tests fail in the standalone-enabled-microprofile-test execution on the nightly Windows JDK 11 jobs.
[22:56:29][org.wildfly:wildfly-ts-integ-microprofile] [ERROR] Errors: [22:56:29][org.wildfly:wildfly-ts-integ-microprofile] [ERROR] MicroProfileConfigConvertersTestCase.org.wildfly.test.integration.microprofile.config.smallrye.converter.MicroProfileConfigConvertersTestCase » Deployment [22:56:29][org.wildfly:wildfly-ts-integ-microprofile] [ERROR] ConfigSourceFromClassTestCase.org.wildfly.test.integration.microprofile.config.smallrye.management.config_source.ConfigSourceFromClassTestCase » Deployment
These both install a custom module that the test uses, and then remove it.
This testsuite module executes the same tests in two different executions, standalone-enabled-microprofile-test and default-test. The default-test execution runs first and the tests pass there. The only difference between the two exections is use of standalone-microprofile.xml vs standalone.xml, and I see no reason that would affect behavior. These also work fine on Linux, so my expectation is this has something to do with file locking or something causing the first execution to pollute the environment that the second execution uses.
Edit: I see this fails on the JDK 8 job too, and on the PR jobs for Windows. TeamCity seems to get confused about running tests twice, in different surefire executions, and doesn't properly report failures. It seems to only be the Windows JDK 11 job where the failure is reported.